New distribution records of two bamboo species in Yunnan, China with description of the inflorescence for Melocalamus yunnanensis (Poaceae, Bambusoideae)

Abstract Ampelocalamus actinotrichus (Merrill & Chun) S. L. Chen, T. H. Wen & G. Y. Sheng and Neomicrocalamus prainii (Gamble) P. C. Keng are reported with new distribution records in southern and southeastern Yunnan, China, respectively. Ampelocalamus actinotrichus was previously recorded to be endemic to Hainan, China, and Neomicrocalamus prainii to be distributed in southern Tibet and western Yunnan in China, northeastern India, and Burma. The identities of individuals collected in southern and southeastern Yunnan of these two species are confirmed by molecular evidence. The new distribution record of Ampelocalamus actinotrichus provides a case at the species level for confirming floristic affinities of southern Yunnan and Hainan Island in south China. The disjunct distribution of Neomicrocalamus prainii in Yunnan is concordant with the ecogeographical diagonal line from northwestern Yunnan to southeastern Yunnan and this may imply a tropical origin of this species. In addition, the inflorescence of Melocalamus yunnanensis (T. H. Wen) T. P. Yi is described.


Introduction
Although morphological descriptions, distributions, and pictures of most bamboo species and genera in China have been recorded in the Flora Reipublicae Popularis Sinicae (FRPS) (Keng and Wang 1996), the Flora of China (Li et al. 2006), and the Iconographia Bambusoidearum Sinicarum (Yi et al. 2008), there are still many aspects that need to be changed incorporating fi ndings from molecular phylogenetic analyses, the re-evaluation of morphological characters, and additional fi eld investigations. During our fi eld work from 2012 to 2014, two distinct bamboo species were noted. One occurs in Xishuangbanna, southern Yunnan, China, which was initially identifi ed as "Ampelocalamus menglaensis" Hsueh & F. Du nom. nud. (Du et al. 1999), but closely resembled A. actinotrichus (Merrill & Chun) S. L. Chen, T. H. Wen & G. Y. Sheng. Th e other is distributed in Malipo, southeastern Yunnan, China, where it is called "teng zhu" (climbing bamboo) by local people, and was tentatively identifi ed as Neomicrocalamus prainii (Gamble) P. C. Keng. Ampelocalamus actinotrichus, the type species of the genus Ampelocalamus S. L. Chen, T. H. Wen & G. Y. Sheng (Chen et al. 1981), belongs to the tribe Arundinarieae (BPG 2012), and is endemic to the island of Hainan, China, according to the FRPS and the Flora of China. Neomicrocalamus prainii is the type species of the genus Neomicrocalamus P. C. Keng which belongs to the subtribe Bambusinae (Bambuseae) (Keng 1983, BPG 2012, and is distributed in southern Tibet and western Yunnan of China (Keng andWang 1996, Li et al. 2006), and northeastern India (Meghalaya and Nagaland) (Ohrnberger 1999, Kumari andSingh 2013). If the two species discovered in southern and southeastern Yunnan, are indeed Ampelocalamus actinotrichus and Neomicrocalamus prainii, respectively, their collections would represent new distribution records in China for those species.
In order to further confi rm the identities of these two species in southern and southeastern Yunnan, molecular phylogenetic analyses were carried out. Previous molecular phylogenetic studies demonstrated that plastid regions had low resolution at species and generic levels in Arundinarieae Clark 2010, Zeng et al. 2010), and conspicuous confl icts between plastid and nuclear phylogenetic tree topologies have been found (Zhang et al. 2012. Moreover, nuclear phylogenies were more congruent with morphology-based classifi cations and had higher resolutions than plastid phylogenies. Th erefore, we selected one nuclear gene LEAFY  for the molecular identifi cation of the "Ampelocalamus menglaensis" sample. Th e standard DNA barcoding plastid regions had been found to possess a certain discriminative power at species and generic levels in the subtribe Bambusinae and can be obtained by direct sequencing, while obtaining nuclear genes may involve cloning which requires more time and resources , Goh et al. 2013. Given their convenience and effi ciency, we selected four plastid regions (rbcL-psaI, rpl32-trnL, trnG-trnT(g), trnG-trnT(t)) that show higher levels of variation than standard DNA barcoding markers , for identifying the "teng zhu" sample.
In addition to the two distinct bamboo species, we collected fl owering specimens of Melocalamus yunnanensis (T. H. Wen) T. P. Yi in the fi eld and the Bamboo Garden of Xishuangbanna Tropical Botanical Garden. Th e infl orescence of this species is described in this paper, and its identity as a member of the genus Melocalamus Bentham reconfi rmed.

Field collections
Specimens for morphological observations and silica gel-dried leaf samples of "Ampelocalamus menglaensis" and Melocalamus yunnanensis were collected in August 2012 and April 2013 in southern Yunnan (Mengla, Xishuangbanna) and of "teng zhu" in October 2013 and August 2014 in southeastern Yunnan. Th e distribution map was created using DIVA-GIS (http://www.diva-gis.org).

Morphological observations
Th e specimens collected in southern Yunnan were compared with specimens of Ampelocalamus actinotrichus from Hainan Island. Th e specimens of "teng zhu" from southeastern Yunnan were compared with those of Neomicrocalamus prainii from northwestern Yunnan and southern Tibet.
Th e infl orescence of Melocalamus yunnanensis from the Bamboo Garden of Xishuangbanna Tropical Botanical Garden was compared with specimens of this species from Jiangcheng, southern Yunnan. Th e glumes, lemma, palea, ovary, style, and stamens were observed under a hand-lens (30×).

Taxon sampling
Th e phylogeny of Arundinarieae based on LEAFY indicated that the genus Ampelocalamus is monophyletic, except for A. calcareus C. D. Chu & C. S. Chao, and sister to Drepanostachyum P. C. Keng and Himalayacalamus P. C. Keng ). Th us, samples of the latter two genera were used as outgroup in our study. We aimed to elucidate the identity of "A. menglaensis", not to infer phylogenetic relationships. Th erefore, only those taxa closely related to A. actinotrichus were included (Table 1).
Voucher specimens of all samples are deposited at the herbarium of the Kunming Institute of Botany, Chinese Academy of Sciences (KUN). Table 1. Plant materials, voucher information, and GenBank accession numbers of the samples used in the phylogenetic analyses.

DNA isolation, amplification, and sequencing
Total genomic DNA was extracted from silica gel-dried leaves using a modifi ed CTAB procedure (Doyle and Doyle 1987). Primers and protocols for the PCR amplifi cation of the nuclear gene LEAFY and four plastid regions (rbcL-psaI, rpl32-trnL, trnG-trnT(g), trnG-trnT(t)) followed prior studies . PCR products were checked on 1% agarose gels, and purifi ed using ExoSAP-IT (USB, Cleveland, OH, USA). Double-stranded and purifi ed PCR products were sequenced by the dideoxy chain termination method with ABI PRISM Bigdye Terminator Cycle Sequencing Ready Reaction Kit with AmpliTaq DNA polymerase FS (Perkin Elmer, Waltham, MA, USA). PCR reactions and programs were chosen according to the recommendations of the handbook, with slight modifi cations in some cases. Bidirectional sequencing was performed on an ABI 3730xl automated sequencer.

Phylogenetic analyses
Sequences were assembled and edited with SeqMan (DNA STAR package; DNA Star Inc., Madison, WI, USA), aligned by MUSCLE (Edgar 2004), and adjusted manually where necessary. Informative indels introduced by the sequence alignment were coded as binary characters using the simple indel coding method of Simmons and Ochoterena (2000). Due to no obvious confl icts among individual plastid trees, we used a combined data set of four plastid regions for phylogenetic analyses. All data matrices are available on request from the fi rst author. Sequences newly obtained in this study have been deposited in GenBank (Table 1). Accession numbers initialed with 'KM' and 'KF' were downloaded directly from GenBank, and the others were obtained in this study. For phylogeny reconstructions, we used three methods, namely maximum parsimony (MP), maximum likelihood (ML), and Bayesian inference (BI). Th e MP and ML analyses were conducted with PAUP* version 4.0b10 (Swoff ord 2002). For the ML analyses, the best-fi tting models were selected using jModeltest v2.1.4 under the Akaike Information Criterion (AIC) (Darriba et al. 2012). Th e TIM3 model and TPM1uf+G model were selected for LEAFY and combined plastid regions, respectively. MP and ML analyses were conducted with the same parameter setting for the heuristic search and the bootstrap calculation. Th e heuristic search was performed with 1000 random addition sequence replicates and TBR branch swapping, MULTREES option in eff ect. Strict consensus trees were calculated for all MP analyses. Branch support was estimated with 1000 bootstrap replicates (Felsenstein 1985) using the heuristic search method as described above (with 100 random addition sequence replicates). Th e same models were used for Bayesian analyses with MrBayes version 3.2.5 (Ronquist et al. 2012). Two independent runs were conducted simultaneously starting with random trees, and each run consisted of one cold and three hot chains. Chains were run for 150000 generations for LEAFY and 200000 generations for the combined plastid data set, and trees sampled every 100 generations. Th e average standard deviation of split frequencies between both runs reached a value below 0.01. Th e convergence of the chains and the number of trees to be discarded were determined using Tracer version 1.6 (http://tree.bio.ed.ac. uk/software/tracer). Th e initial 25% trees were discarded as burn-in, and the remaining trees were used to construct a 50% majority-rule consensus trees. MP and ML bootstrap values ≥ 50% and BI posterior probabilities ≥ 0.95 were labeled on the tree branches.

Phylogenetic analysis
Th e aligned length of LEAFY was 738 bp, and six indels were coded as additional absent/ present (0/1) characters, giving a total of 744 characters in the MP matrix, of which 26 were parsimony-informative. Sequences of the four plastid regions were obtained for all samples, except trnG-trnT(t) for the sample ZXZ11027. Th e combined plastid matrix was 3737 bp long and included 14 indel characters in the MP matrix, 113 of which were parsimonyinformative. Th ere were inversions in the rbcL-psaI and rpl32-trnL sequences (Zeng et al. 2010) and gaps were introduced to separate the inverted regions to avoid overweighting the inversions. Th ose gaps were not scored and treated as missing data.
Th e 50% majority-rule consensus tree from BI for LEAFY is presented in Fig. 1. Th e ML tree and the MP strict consensus tree were consistent with the Bayesian tree except for a few branches with very low support (not shown). Th e statistical support was shown along the branches (MP/ML/BI). All individuals of Ampelocalamus formed a clade with high support (MP/ML/BI = 98/98/1.00). Th e samples of "A. menglaensis" (voucher numbers 12167, 13007, and 13012) fell with the Ampelocalamus actinotrichus samples collected from Hainan.

Ampelocalamus actinotrichus in southern Yunnan and its biogeographic implications
Our collections (specimens 12167, 13007, and 13012) were identifi ed as Ampelocalamus menglaensis nom. nud. by bamboo taxonomists. Th is species was initially published without proper description and designation of the type specimen, and it was reported that this bamboo was originally distributed in Mengla, Yunnan, China (Du et al. 1999). In the Flora Yunnanica (Sun et al. 2003), it was suggested that this species could represent a misidentifi cation of Dinochloa puberula McClure and was not included in the genus Ampelocalamus. Later, Wang (2004) described this species briefl y in Chinese with two colored photos in his book Ornamental Bamboos. Based on the brief description and our observations in the fi eld, we inferred that this species should be Ampelocalamus actinotrichus. Major morphological diagnostic characters included: culms pendulous or scrambling; rhizomes pachymorph; internodes with sparsely brown setae initially; nodal sheath scars prominent; branches two to several, occasionally with one dominant branch replacing the main culm and scrambling to other plants; culm leaves tardily deciduous, much shorter than internodes, sheaths with sparse setae abaxially, auricles ovate, oral setae radiate and initially purple, margins of ligules with long and purple hairs, blades lanceolate; leaf sheaths glabrous or with sparse setae abaxially, auricles kidney-shaped, oral setae radiate and initially purple, ligules fi mbriate, blades with long pubescence on both epidermises (Fig. 4). Yang (2005) analyzed the anatomy of leaves and roots of 13 species of Drepanostachyum P.C. Keng and Ampelocalamus, including A. actinotrichus and A. menglaensis nom. nud. Ampelocalamus actinotrichus and A. menglaensis nom. nud had similar features but distinct from other species, such as two layers of mesophyll cells in leaf transverse sections, large fusoid cells, fewer prickles on abaxial and adaxial epidermises, microhairs developed. Moreover, our molecular phylogenetic analyses demonstrated that our collections from Yunnan grouped together with Ampelocalamus actinotrichus from Hainan, and there was a little genetic divergence between these two biogeographic entities (Fig. 1). On grounds of the aforementioned evidence we can confi rm that Ampelocalamus menglaensis nom. nud. is conspecifi c with A. actinotrichus.
Ampelocalamus actinotrichus was recorded to occur only on the island of Hainan before our confi rmation of its occurence in southern Yunnan (Fig. 5). During our fi eld investigations in Mengla, Xishuangbanna, southern Yunnan we went to the border of China and Laos (collecting voucher specimens 13012 and 13013; 13013 was not used here for the molecular analyses). We were told by the local people that this bamboo could also be found in northern Laos. Chen and Wen (2008) reported that this bamboo is distributed in tropical montane rainforests of Mengsong, southern Yunnan, which is adjacent to Burma. It suggests that Ampelocalamus actinotrichus may be distributed in Burma. Additional fi eld explorations are needed to clarify whether this species is more widely distributed on the Indo-China peninsula. Th e new distribution record of this species has a great signifi cance in resolving the origin and divergence of the genus Ampelocalamus.
Mengla and Mengsong are both parts of Xishuangbanna, in southern Yunnan. Th e fl ora of Xishuangbanna is part of the Indo-Malesian fl ora, and has a close affi nity with fl oras of adjacent areas (i.e., southern China including tropical Guangxi and Hainan, Burma, Laos, Th ailand, and Vietnam) (Zhu 1994, Zhu et al. 2001, Zhu and Roos 2004. Th e distribution of Ampelocalamus actinotrichus in both Hainan and southern Yunnan (and likely in Burma and Laos) provides a case at the species level for confi rming the affi nity of fl oras across southern China including Xishuangbanna (and likely adjacent areas).

Neomicrocalamus prainii in southeastern Yunnan and its biogeographic implications
From the point of view of morphology, individuals of "teng zhu" (specimens 13045, YXY150, YXY151) from southeastern Yunnan share many features of Neomicrocalamus prainii, including culms scrambling, nearly solid, branches many with the dominant branch equal in size to the culm and other small branches seldom branching again, culm leaf sheaths with purple-brown spots and white pubescence abaxially, culm leaf blades acicular, and others (Fig. 6). Th e habitats of "teng zhu" in southeastern Yunnan are mainly rocky mountains which are similar to those we found in Gongshan, northwestern Yunnan (voucher No. ZXZ11027). Th e molecular phylogenetic analyses illustrated that those individuals fell with Neomicrocalamus prainii from southern Tibet and northwestern Yunnan, especially close to the individual from northwestern Yunnan (voucher No. ZXZ11027) (Fig. 2). Based on molecular and morphological evidence, we concluded that "teng zhu" should be identifi ed as Neomicrocalamus prainii.  (Li 1994). Li (1994) inferred that an ecogeographical diagonal line from northwestern Yunnan to southeastern Yunnan was caused by a northward and clockwise rotation of the Shan-Malay Plate (Burma-Malaya Geoblock) since the Miocene, and that northwestern and southeastern Yunnan were once at the same latitude both with a tropical environment. One of the biological eff ects of this plate movement was that some species were discontinuously distributed at both ends of the diagonal and some were concentrated on the southwestern side (Li et al. 1999, Zhu andYan 2002). Th e disjunct distribution of Neomicrocalamus prainii in northwestern and southeastern Yunnan are concordant with this biogeographical line (Fig. 5). Neomicrocalamus prainii is distributed at both ends of the diagonal. Th e distribution pattern of N. prainii may imply that this species has a tropical origin, and it provides another example for verifying the reality of the ecogeographical diagonal line in Yunnan.

Melocalamus yunnanensis
Th is species was initially described as Racemobambos yunnanensis based on incomplete, poor specimens (Wen 1986). However, it is easily identifi ed by observing the peculiar culm leaves. It was transferred to the genus Neomicrocalamus (Ohrnberger 1999), and subsequently subsumed into Melocalamus (Yi et al. 2007). Molecular studies also showed that it was not a member of Racemobambos or Neomicrocalamus and had a close relationship with Melocalamus . Main infl orescence features of Melocalamus include pseudospikelets 2-fl owered, glumes 2, palea 2-keeled, lodicules 3, stamens 6, stigmas 2 or 3, plumose. Melocalamus yunnanensis as a member of the genus Melocalamus was confi rmed by infl orescence characters reported in the current paper.